1. Decoupling of Iron and Phosphate in the Global Ocean
Payal Parekh
MIT/WHOI Joint Program
April 18, 2003
In collaboration with Mick Follows, Ed Boyle and John Marshall

2. Thesis Aims Develop a mechanistic model of iron cycling to:
reproduce observed [Fe]
predict rates of biogeochemical processes
understand controls on Fe in upwelled waters
gain insight on decoupling of Fe and PO4
Motivation: Understand iron’ s role in possible pCO2 drawdown.

3. High Nutrient, Low Chlorophyll Regions
Latitude µM
Longitude
Conkright et al. (1994)

4. Chlorophyll Distribution
SeaWiFS image

5. Could Lack of Iron be the Culprit?
Sources of Iron to surface waters:
Upwelling
Aeolian Dust Flux
mg Fe m -2 yr -1
Gao et al. (2001)

6. Link between dust flux and pCO2?
Atmospheric CO2 (ppm)
Dust Flux (mg m -2 yr -1)
Age (kyr)

7. Iron Addition Experiment
SOIREE – S. Ocean

8. Why are Fe and PO4 decoupled?
R=Fe:PO4 biological uptake ratio

9. Iron Biogeochemistry Figure from Bergquist DUST surface
biological
loop
dissolved Fe
(< 0.4 mm)
biogenic export
lateral transport
and mixing
DUST
refractory dust
upwelling and
vertical mixing
mixed layer bottom
surface
remineralization
sediment-water interface
scavenging
& desorption
mixing
sedimentary deposition
Fe’ + L’  FeL
Figure from Bergquist

10. Iron Observations 1000m Surface
Compiled from the literature and Boyle (unpub).

11. Is it possible to reproduce and understand controls on observed deep water Fe distributions?
Perform sensitivity tests to constrain rates of biogeochemical processes using an idealized multi box model
Test iron parameterization in the MITGCM, resolve intrabasinal differences and vertical profile of iron

12. Box Model Simulations
Volume transport chosen to fit Radiocarbon, Broecker/Peng (86,87)
Tracers: PO4, DOP, Fe
Biological Uptake: µ*[Fe/(Fe+Ks)]
Michaelis-Menten kinetics – Fe, light limited

13. Iron Parameterization
K= binding strength of ligand
Log(K) ranges
FeL=K[Fe’][L’]
LT=L’+FeL
LT = 1 nM, based on Rue/Bruland (1995) and others

14. Results: Complexation Case
0.6 nM
Atlantic
0.3 nM
S. Ocean
0.4 nM
Indo-Pac.
log K
Ksc yr-1

15. Summary: Box Model Results
Iron parameterization successfully reproduces deep water Fe gradients
Sensitivity study constrains scavenging rate over wide range of K (ligand binding strengths).

16. Explorations with an Ocean General Circulation Model
Same parameterization as in box model
Forced seasonally with Gao et al. (2001) dust flux estimate

17. Modeled Fe
Surface
935m
Fe exceeds solubility (Liu and Millero, 2002) in surface waters
Successfully reproduces observed Fe gradients at depth

18. Observations: PO4
Surface
1000m
Conkright et al. (1994)

19. Modeled PO4
Surface
1000m
Fe limitation term [µ*(Fe/Fe+Ks)] successfully reproduces HNLC region
Previously, models either restored to surface or adjusted export time scale
Model has difficulty in subtropical Pacific surface waters (nutrient trapping)
Deep water PO4 agrees well with observations

20. Fe*: Measure of decoupling between Fe and PO4
Fe* = Fe-RPO4
Interpretation of Fe*:
Positive Fe*: Adequate Fe to support biological utilization of PO4
Negative Fe*: Deficit in Fe
Fe:PO4 ratio (R) must be specified

21. Fe*: Zonally averaged section in the Atlantic
Calculated residence time for GCM: 285 yr
Positive at surface in northern hemisphere
As NADW travels southward, scavenging depletes Fe*
AAIW/AABW both negative due to low aeolian dust flux in S. Ocean

22. Fe*: Indo-Pacific Positive Fe* at surface between 25-40° N
Decoupling of Fe and PO4 greatest in deep, old waters of N. Pacific.

23. North Pacific vs. North Atlantic
Aeolian Fe flux (mg Fe m-2 yr-1)
Both basins have high dust flux and are regions of upwelling
Why is the N. Pacific Fe limited?

24. Fe* below the mixed layer
Fe* at 290m
Aeolian dust flux is not enough to compensate for the low Fe, high PO4 waters of the old, upwelled waters in the N. Pacific

25. GCM Simulations: Summary
Mechanistic parameterization of Fe that includes complexation and scavenging reproduces deepwater Fe gradients
The model exceeds solubility (Liu and Millero, 2002) at the surface in high dust flux regions
Fe* is a useful tracer for understanding how/why Fe and PO4 become decoupled in HNLC regions

26. Utility of mechanistic Fe model
A mechanistic Fe model allows for exploration of important climate questions, such as Does increased dust flux lead to increased efficiency of the biological pump and subsequent drawdown in pCO2?

27. Surface PO4 Sensitivity to Increased Dust Flux: Preliminary Results
‘Paleo’ dust estimate from Mahowald et al. (1999)
Dust flux greater nine times globally
63 fold increase in the Southern Ocean

28. Southern Ocean Surface PO4 Response
Drawdown of ~ 0.5 µM
Due to low surface PO4, results should be viewed as a lower bound estimate
Watson et al. (2000) estimated 0.6 µM drawdown of PO4 - ~ 30 ppm drawdown
PO4
(µM)
Δ PO4

29. Thesis Conclusions
Developed a mechanistic model of Iron cycling for the global ocean:
Sensitivity study using box model suggests iron scavenging rate ranges between yr-1 log(K) between 10-13
Complexation and scavenging description successfully reproduces deep water Fe gradients in the GCM
Fe* is a useful tracer for understanding the decoupling of Fe and PO4
Dust flux is not enough to compensate the low Fe* of upwelled waters in HNLC regions
A nine fold increase in dust flux results in ~ 0.5 µM drawdown of PO4

30. Research Recommendations
A pressing need for Fe measurements, especially in the deep waters
Identification of source(s), sink(s) and chemical characterization of Fe-binding ligands
Experiments aimed at studying processes affecting Fe in surface waters

31. Works in Progress
Improving surface iron distribution in model– in collaboration with M. Follows, E. Boyle
Coupling iron model with a more sophisticated ecosystem model – in collaboration with S. Dutkiewicz, M. Follows
Exploring more closely iron cycling in the S. Ocean – in collaboration with T. Ito, J. Marshall

32. Acknowledgements J. Marshall for inviting me to join his group
M. Follows for being a patient and encouraging advisor
E. Boyle for sharing his insights on iron chemistry and the intricacies of making iron measurements
J. Moffett for enlightening me on iron’s role in ocean ecology
M. Bacon for pointing me to the parallels between iron and thorium
T. Voelker for acting like a committee member and being a supportive female faculty member of the Joint Program

33. Acknowledgements
S. Dutkiewicz and T. Ito for many stimulating discussions on ocean biogeochemistry
J. Campin for teaching me how to program (and debug) in FORTRAN
C. Hill for introducing me to the joys of running on parallel computers
M. Losch and D. Ferreira for sharing with me their encyclopedic knowledge of MATLAB
The folks on the upper floors of the Green Building for their friendship and assistance at various times for various tasks

34. Acknowledgements
Thesis Crisis Intervention Committee: Bridget Bergquist, Mick Follows, Chris Hill, Taka Ito, Alex Rae, Steph Dutkiewicz , Ariane Verdy, Fanny Monteiro, Oren Weinrib
Friends that fed and housed me during the dark days of thesis writing: Anke, Jen and Pedro, Fernanda, Nico, Anne, Vijay family, Patrick, Ariane, Fanny
To the many friends (old and new) I have made at MIT, WHOI and around the city for reminding me that there is a world beyond the Green building
The MIT/WHOI Joint Program in Oceanography
To my family for their love and support

35. Funding Acknowledgement
NASA Global Change Fellowship

36. Scavenging/Desorption Case
Thorium, a metal with similar properties to Fe, is known to desorb from particles (Bacon and Anderson, 1982).
Perhaps Fe is also involved in this process.

37. Scav/Desorption Results
Atlantic
S. Ocean
kb yr-1)
Indo-Pac
Scav. Rate (yr-1)

38. Observations: Iron Surface 1000m Below 2500m
Compiled from literature, unpub. results from E. Boyle

39. GCM Simulations
In the context of the more sophisticated physical description of the MITGCM, net scavenging and desorption case are not adequate to describe the iron cycle
Complexation case agrees well with measurements below the surface
explains role of transport and scavenging in decoupling Fe and PO4

40. Iron Observations: Surface
Compiled from literature and Boyle (unpub.)

41. Iron Observations: At depth
1000m
Z> 2500m
Compiled from literature, Boyle (unpub.)

42. Net Scavenging Case Simplest Description
Net scavenging term (K_sc) is the sum of the various geochemical processes Fe is involved in
Biological export is Fe, light limited

43. Results: Scavenging Case
For < knetsc < 0.006, observed gradients are produced
Description does not resolve various biogeochemical processes

44. Iron Biogeochemistry
Figure courtesy of B. Bergquist

45. Comparison of observed and modeled profiles
Pacific: 3°S, 140°W
Atlantic:10°N, 45°W
Boyle et al. (unpub.)
Johnson et al. (1997)